Energy consumption, energy efficiency, and consumer perceptions: A case study for the Southeast United States

Applied Energy 165 (2016) 660–669

Contents lists available at ScienceDirect

Applied Energy journal homepage: www.elsevier.com/locate/apenergy

Energy consumption, energy efficiency, and consumer perceptions: A case study for the Southeast United States Christopher A. Craig ⇑ Environmental Dynamics, University of Arkansas, United States

h i g h l i g h t s  Interaction between climate, efficiency, and electricity consumption were examined.  2450 state residents were surveyed about clean energy and subsidy policies.  Indirect energy efficiency costs negatively influenced electricity consumption.  Cooling degree days were positively related to electricity consumption.  Resident awareness influenced policy perceptions about clean energy and subsidies.

a r t i c l e

i n f o

Article history: Received 17 August 2015 Received in revised form 15 December 2015 Accepted 17 December 2015

Keywords: Energy efficiency Theory of planned behavior Heating degree day Cooling degree day Electricity Climate variability

a b s t r a c t This study examined the interaction between climatic variability and residential electricity consumption in a Southeast US state. Residential electricity consumers were surveyed to better understand how to diffuse positive attitudes and behaviors related to energy efficiency (EE) into households. The study found that 16.8% of the variability in residential electricity consumption for heating applications was explained by indirect EE costs. 36.6% of the variability in residential electricity consumption for cooling applications was explained by indirect EE costs and cooling degree days (CDD). A survey of 2450 residential electricity consumers was analyzed using the theory of planned behavior (TPB). Significant findings suggest that those residents are aware of utility EE programs are more likely to participate, view utility company motives more favorably, to support governmental subsidies for EE programs, and to support the use of clean energy by utility companies. Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction With the understanding that the energy consumptive patterns in the United States (US) are a contributing factor to anthropogenic climate change [1], this study seeks to gain a comprehensive understanding of the relationship between energy consumption, energy efficiency (EE), climate variability, and residential electricity consumer perceptions in the Southeast United States (US). According to the US Energy Information Administration [2], the US is among the highest per capita consumers of electricity in the world, using approximately four times as much electricity as the most consumptive country in the world, China. Carbon emissions continues to rise at historic rates, with emissions more than doubling since 1986 [3]. According to Heede [3], emissions are largely driven by fossil fuel and cement producers, with only

⇑ Tel.: +1 479 466 8505. E-mail address: [email protected] http://dx.doi.org/10.1016/j.apenergy.2015.12.069 0306-2619/Ó 2015 Elsevier Ltd. All rights reserved.

90 such companies responsible for over 60% of global carbon emissions since the Industrial Revolution. As the largest electricity consuming sector, particularly in the Southeast US where states are more reliant on fossil fuels and per capita usage is higher than other regions in the US [4], residential consumers are a salient driver of carbon emissions related to the production of electricity. In order to ensure continued, secure energy access and lowered reliance on carbon rich fossil fuel sources, short- and long-term regulatory practices are needed to achieve production and emissions goals in the energy markets [5]. The evaluation of energy mix is of great concern. Both in the US and in other industrialized countries globally [3,6], fossil fuel reliant energy producers continue to contribute GHG emissions at higher rates than other groups. While the percentage of fossil fuels in the US and abroad in terms of percentage energy mix has decreased [7,8], issues such as increased electricity demand from non-traditional users (e.g., transportation), increased economic activity, population growth, and energy security have resulted in

C.A. Craig / Applied Energy 165 (2016) 660–669

increased consumption and continued reliance on fossil fuels [8–12]. A host of technologies are available to reduce GHG emissions beyond those traditionally deployed with varying degrees of cost-effectiveness (e.g., [13–15]). However, there has been some reluctance among residents around the world to embrace clean energy sources and efficiencies in their own homes largely due to lack of awareness [16–18]. To further policy and practice, particularly around cost-effective methods to reduce consumption and emissions, the engagement of residential energy users is crucial. Residential energy use is expected to increase carbon emissions for the sector from 17% to 21% in the US by the year 2020 [19], magnifying the implications of rising emission levels relative to energy producers. According to Shove [20], ‘‘the challenges of climate change are such that many familiar ways of life and many of the patterns of consumption associated with them are fundamentally unsustainable” (p. 1273). There are positive feedbacks in the consumptive US electricity system. Increased consumption leads to increased GHG emissions which has been shown to influence climatic variability and extreme weather events [1]. To help reduce energy consumption and related GHG emissions, Fisher and Newell [21] suggest that both policy and the diffusion of relevant knowledge through effective communication as to influence positive behavior is necessary. The current study seeks to expand beyond merely identifying energy related problems in an effort to understand the mechanisms by which EE can be diffused directly into households. 1.1. Energy efficiency programs and climate Energy related decisions to curb consumption, ranging from federal energy policy to the type of light bulb in the home, are people-centric. To help slow energy consumption and the related GHG emissions in the US, governmental agencies as well as investor-owned, state-regulated utility companies engage in EE programs to influence adoption of technologies and proconservation behaviors [18]. There are billions in incentive dollars available from utilities and governmental agencies for residences to become more efficient [22], with over 30 million US dollars deployed in the focal state in 2012. The deployment of incentives to those who utilize these programs is largely based on a deemed savings model, in that efficiency upgrades are assigned a kilowatt– hour (kW h; unit of measurement of electricity) savings value approved by a state regulatory body [18]. Relying on these assigned values instead of using pre- and post-test consumption analysis make it difficult to gauge the true impact of such programs. Because of these complications, the current study will focus on actual peak electricity kW savings reported by utility companies in lieu of deemed kW h household savings. Also, the study will focus only on indirect EE costs that include non-incentive spending such as marketing and administration, as direct costs are incentives paid based on the deemed kW h values. Electricity consumption and electricity savings from EE programs were reported by utility companies with the EIA. However, there is no systematic control for climatic factors in these reports. In a longitudinal residential study, Jovanovic et al. [23] demonstrated that temperature was the biggest determinant for increased electricity consumption, particularly during periods of extreme cold and hot temperatures related to electric heating and cooling equipment. Large empirical studies indeed demonstrated that both electricity (r = .84; [24]) and natural gas (r P .97; [25]) consumption are strongly linked to climatic factors such as heating degree days (HDD) and cooling degree days (CDD). HDD and CDD are measures of how much energy is needed to heat or cool a facility given local temperature conditions, where ‘‘A degree day indicates that the daily average outdoor temperature

661

was one degree higher or lower than some comfortable baseline temperature” [26]. According to Mourshed [27], HDD and CDD are more reliable measures of climatic impact on energy consumption than temperature alone, thus they were included as the measures of climatic variability in this study. Models predict increased temperature variability, including increased electricity demand associated with CDDs absent other factors ([1,28,29]). A salient factor not included in the models is efficiency [28]. The pricing for residential customers is traditionally volumetric, meaning that as demand increases for electricity, residential pricing stays the same [30]. With the Southeast US projected to experience more weather extremes and climatic variability associated with increasing temperature [31], the deployment of effective efficiency programs to offset the projected demand in electricity [28] in the residential sector without the option of variable pricing is crucial. Efficiency programs can range from purchasing discounted efficient lighting at major retailers to making home retrofits [18], with the entire portfolio of electricity savings measures needed to combat increased demand [32]. The current study examines the influence of EE programs (i.e., actual kW savings and costs of programs), HDD, and CDD on kW h consumption per consumer in Southeast US. More specifically, the study examined these relationships primarily relative to electric heating applications and electric cooling applications: Research Question 1. How much variability in residential kW h consumption used for heating and cooling is explained by climatic factors, EE program actual kW savings, and EE costs? 1.2. Communication and the residential electricity consumer Communication with electricity consumers is essential to ensure that energy savings occur. For instance, Delmas et al. [33] found in a meta-analysis that incentive programs administered without feedback mechanisms resulted in increased energy consumption in the home, the opposite of the desired effect. To combat results in the wrong direction, or the rebound effect [34], states are increasingly using feedback rich deep-savings approaches that behaviorally empower residential customers to reduce electricity consumption. Asensio and Delmas [35] saw consumption reductions when this strategy was used with residential electricity customers. Darby [36] demonstrated that rich feedback can behaviorally lead to energy savings between 5% and 15%, whereas behavioral reduction in consumption outside of feedback is minimal. Craig and Allen [37] had similar results, in that households saw a year-over-year drop of over 10% in electricity consumption after a behavioral intervention that included rich feedback when controlling for climatic variability. While there are some in the US that are deploying aggressive behavioral programs (e.g., O’Power, the Shelton Group), pro-active behavioral interventions in residences remain the exception. It is not as easy as just providing incentives or presenting a message related to participating in EE and expecting people to change, however. Awareness about efficiency and related programs remains low among adults and children [18,37]. For instance, in a recent study, only 21% of residences interviewed recalled receiving information or educational materials about efficiency [19]. Dewaters and Powers [38] noted that energy literacy has an affective, or emotional, element. Mis-information and previously formed attitudes have the potential to deter the receipt of new information and further solidify potentially negative attitudes that can deter positive behaviors. In fact, Craig and Allen (2014) found individuals who did not know about utility EE programs were less supportive of the use of alternative energy, which has the potential

662

C.A. Craig / Applied Energy 165 (2016) 660–669

to further hinder the development of non-carbon emitting infrastructure. Also, when considering EE and other pro-conservation actions, the gap between positive attitudes and perceptions and the actual behavior is well-documented (e.g., [39–42]). This gap highlights the need to use effective communication and messaging to build knowledge and positive attitudes that increase the likelihood of pro-conservation behaviors. The theory of planned behavior [43] states that people become aware of a topic, form attitudes and perceptions about the topic, and plan to behave accordingly. When dealing with conservation related behaviors, the TPB is complicated by low awareness levels about EE and the gap between perceptions and behavior. Engagement in socially responsible programs such as efficiency can influence individual attitudes and affect [44], increasing the likelihood of organizationally desired behaviors such as residential participation in efficiency programs. Micro-level engagement by residences can also increase the participation of others in socially responsible and/or environmental initiatives by providing normative pressures or nudges [45,46]. In organizational settings, normative social pressures in both theory and practice are related to proconservation behaviors such as energy consumption [47,48]. Differences in environmental attitudes and behaviors have emerged in the past, however, on demographic factors including age, gender, political party affiliation, and income (e.g., [18,49–51]). Individuals are paying into efficiency programs in the form of riders or fees on their electric bills, but the majority of residential ratepayers remain unaware about programs and do not participate [18,19]. Of interest to utility organizations pursuing consumptive behavior changes, individuals not aware of environmental actions by an organization perceived actions less favorably than those who were aware [44]. Consistent with these findings, Liu et al. [17] found that residents who were involved prior to efficiency upgrades were more likely to realize higher energy savings once a retrofit occurred. In this study, residents were asked if they were aware of rebates for efficient lighting or if they had purchased rebated lighting as part of utility organization efficiency programs. Because of the historically low awareness and participation rates in residential efficiency programs, it was necessary to focus on the most wide-spread program to increase the likelihood that the resident had participated [18,19]. For investor-owned utility companies in particular, the consumer is a very important part of the organization, as consumptive energy patterns directly influence emissions related to production. Participation by residential energy consumers for utility companies is a cost-effective means to reduce consumption and emissions when faced with the increasing need for expensive energy infrastructure as well as regulations to aggressively reduce emission such as the Clean Power Plan recently proposed by the United States Environmental Protection Agency (EPA) [52]. If utility organizations are able to increase awareness and participation in entry-level programs, such as rebates for efficient lighting that are provide more electricity savings per dollar spent than more robust measures [53], and provide feedback to the residential end-user, the likelihood of successful program deployment and savings persistence increases [17,18,35,37]. With a better understanding of the interaction of energy consumption with the climate and EE programs in the focal state, the current study also seeks to understand what differences among residential electricity consumers are driving participation in efficiency programs, influencing residential consumer perceptions about utility companies, and influencing residential consumer support for efficiency programs and clean energy. The current study examines the differences among residential energy consumers in terms of perceived utility motives for efficiency programs, support for government subsidies for EE, and support for utility company use of clean energy controlling for awareness

levels, past participation in utility programs, and demographic factors. The basic framework of the TPB is utilized, exploring how residential awareness and participation in programs can influence attitudes and perceptions, which in turn can influence residential electricity consumer support for clean energy and government EE subsidies. The following research questions are proposed: Research Question 2a. Is there a difference between residential electricity consumers participation in EE programs based on residential electricity consumer awareness levels? Research Question 2b. Is there a difference in perceptions about utility motives for providing efficiency programs, support for government policy for efficiency programs, and support for utility use of clean energy based on residential electricity consumer awareness levels, participation in utility programs, and demographic factors? 2. Materials and methods 2.1. Procedure HDD and CDD climatic data were obtained for the focal state from 1992 through 2012 using NCAR command language version 6.3.0 (2015). HDD and CDD were annualized to examine climatic trends related to temperature variability from 1992 until 2012. Energy consumption and efficiency savings were calculated from data retrieved for the years 1992 through 2012 from Form EIA 861 and the EIA website (http://www.eia.gov/electricity/data/ eia861/). Energy consumption was calculated as kW h per consumer. Efficiency savings were calculated as the actual peak kW savings per consumer in terms of incremental savings (i.e., within year kW savings from new EE participants) and annual savings (i.e., the lifetime kW savings from all EE participants in current and past years). EE costs were calculated as indirect cost per consumer, and included costs not directly related to implementing EE programs such as marketing, measurement and evaluation, and administration. Using a stratified random sampling technique [54], a phone survey was administered in June 2012 to residential electricity consumers in a single Southeastern US state in four counties among registered voters. Phone numbers included both wired and cell phones, depending on the number used for voter registration. Rural/urban, income, party, and race were the socio-economic factors used for county selection. A phone survey was used for two reasons. First, while over 87% of the US population is connected to the internet today (http://www.internetworldstats.com), many more rural states, such as the focal state, have over 30% of their residents who do not use the internet according to the 2009 US Census. Second, Xing and Handy [55] noted that while there was a difference between internet and phone results on demographic questions, there was no significant difference for attitudinal and behavioral questions. The commercial polling firm provided preverified demographic data, making the phone survey suitable for the research. According to PewResearch Center for the People & the Press [56], typical response rates for phone surveys are 9%. The current study had a response rate of 10.7% (n = 2450), and it took each respondent approximately 10 min to answer 12 questions (age, party, and income were pre-verified and provided). The commercial polling firm completed the 2450 surveys from a pre-selected group of 500,000 potential residential electricity consumer respondents. The large sample size increased the internal validity and decreased the margin of error [54]. The number of respondents using the stratified random sampling method returned better than a 5% confidence interval at the 95% confidence level for the sample (http://www.surveysystem.comfsscalc.htm).

C.A. Craig / Applied Energy 165 (2016) 660–669

2.2. Sample Concerning gender, 35% of the respondents were male (n = 416) and 65% of the respondents were female (n = 774). In terms of race, respondents were 74.1% (n = 859) White, 19.3% (n = 224) Black, .6% (n = 7) Hispanic, and 6% Other. 76.3% of the respondents were Democrat, and 23.7% were Republican. 19.1% (n = 468) of the respondents were in the lowest income bracket from 0 to 19 thousand dollars, and the second largest income bracket (100–124 thousand) included 11.3% of respondents (n = 277). Ages ranged from 20 to 99 and was older (Mean = 67.5, SD = 15.9, n = 2450). 24.4% (n = 599) of the respondents were aware of utility efficiency programs, 41.1% (n = 1008) were unaware of the programs, and 34.4% (n = 843) were mis-informed about the programs (i.e., answer that programs did not exists when in fact programs were available). 20.6% (n = 426) had participated in utility efficiency programs, 63.5% had not (n = 1311), and 15.8% (n = 327) were unsure. According to the United States Census Bureau [57], race percentages were approximately the same for the focal state as the sample, average income per household in the focal state fell within the average income bracket for the sample (50–59 k), and there were approximately 15% more women respondents for the sample. 2.3. Measures A full list of energy consumption, EE savings, climate, and EE costs variables are provided in Table 1. Single-item Likert type questions traditionally utilized in political polling have recently been successfully deployed in academics and industry alike to examine environmental related topics [18,51,58]. The proposed survey consisted of 12 Likert-type questions. Because the commercial polling company has previously validated data for age, party affiliation, and income, it was not necessary to include these items in the phone survey. Compact florescent light (CFL) bulbs are recognizable to consumers due in part to wide-scale availability and point-of-sale placement efforts by retailers and manufacturers in the US. The focal utility efficiency program included on the survey instrument was discounted CFLs because of the entry-level nature of CFLs, the relative affordability Table 1 Descriptives for electricity, climate, and costs variables. Variable

N

Min

Max

Mean

SD

Total kW h0 Heating kW h0 Heating kW h1 Cooling kW h0 Cooling kW h1 CDD0 HDD0 EE IA kW h0 EE AA kW h0 EE AA kW h1 Indirect costs0 Indirect costs1 Indirect costs2

21 21 21 21 21 21 21 21 21 21 21 21 21

10,266.00 1606.97 110.03 1597.23 109.36 789.59 1515.21 .00 .00 2.89 .00 .00 .82

14,538.00 2284.50 254.79 2270.65 253.24 1,302.44 2137.62 65.60 3.80 .44 9.45 4.53 3.65

12,581.38 1977.07 24.09 1965.08 23.93 1,042.16 1,854.66 6.93 .6810 .16 1.10 .45 .22

1,070.32 168.19 95.57 167.17 94.99 134.76 153.87 13.70 1.12 .68 2.42 1.07 .87

Note: The differencing [63] detrending method and the equation from Yaffee and McGee [64] were utilized. Subscript values are as follows: 0 = no differencing, actual values; 1 = first degree of differencing; 2 = second degree of differencing. kW h figures are for overall total kW h consumption per consumer, kW h consumption allocated to heating per consumer, kW h consumption allocated to cooling per consumer, and kW h consumption allocated to all other electric applications per consumer. Heating degree day (HDD) and cooling degree day (CDD) experienced a slight increase in variation, so the original values were used. EE IA kW h refers to actual energy efficiency program kW h savings per consumer for incremental (with-in year) reporting, and EE AA kW h refers to actual energy efficiency program kW h savings per consumer for annual (lifetime) reporting. Indirect costs are periphery costs such as administrative or marketing per consumer, and do not include direct and/or incentive costs of programs.

663

compared to more capital intensive efficiency upgrades such as an energy audit, and the increased likelihood that residential electricity consumers may have recently purchased CFLs. The question that gauged awareness had three response categories including (1) yes, (2) no, (3) don’t know: ‘‘Does your electric utility provider offer discounts or coupons that you can use to buy energy efficient compact florescent light bulbs?” The question that gauged participation had three response categories including (1) yes, (2) no, (3) unsure: ‘‘Have you used discounts or coupons from your electric utility provider when purchasing energy efficient compact florescent light bulbs?” Two questions gauged residential electricity consumer perceptions about utility company motives for offering energy efficiency programs, both with Likert-type response categories from (1) strongly agree to (5) strongly disagree: ‘‘My electric utility provider offers discounts or coupons to purchases energy efficient compact florescent light bulbs because they don’t want me to waste money on my bill,” and ‘‘My electric utility provider offers discounts or coupons to purchase energy efficient compact florescent light bulbs because they want to help me save money on my bill.” Two questions were asked to gauge residential electricity consumer perceptions about support for government subsidies and utility use of clean energy sources, both with Likert-type response categories from (1) strongly agree to (5) strongly disagree: ‘‘I believe that it is okay for the state or federal government to subsidize the cost of energy efficiency programs that utility companies provide,” and ‘‘I believe that utility companies should use more clean or alternative forms of energy.” 2.4. Statistical analysis To prepare the electricity data for analysis, national residential energy usage survey figures [59,60] were used to allocate kW h consumption for electric heating and cooling applications. In the focal state, 15.71% of the electricity consumed was for electric heating and 15.62% for cooling, with 68.67% for all other household electricity uses. In the focal state, 62.86% of households used electricity as a heating fuel source. To calculate the total percentage of electricity consumed for heating applications, this value was multiplied by the percentage of residential energy used for heating (i.e., 25.00%). The same procedure was used for cooling, where 98% of residents used electricity as the fuel source for air conditioning, and 16.00% of the total energy consumed yearly in the state was for air conditioning. To determine the kW h allocations for heating and cooling, the following formulas were used: 62.86% households used electricity as a heating fuel source X 25.00% overall energy was used for heating = 15.71% of overall electricity consumed was for heating, 98.00% households used electricity as an air conditioning fuel source X 16.00% overall energy was used for cooling = 15.62% of overall electricity consumed was for cooling. According to the National Appliance Energy Act of 1987 [61] non-electric furnaces must at minimum meet a 78% efficiency standard in the US. However, older equipment still in use is between 56% and 72% efficient [62]. Consistent with industry practice, a conservative 80% efficiency rating was applied to nonelectric heating applications. A 100% efficiency assumption consistent with industry practice was utilized for electric heating and all cooling applications. The 62.86% value for electric heating equipment was discounted from an original value of 69% electric heating equipment used in the focal state to account for the additional 6.14% of total heating energy loss due the 20% inefficiency. The climatic and energy data were graphically examined to identify trends. The data were detrended using a differencing

664

C.A. Craig / Applied Energy 165 (2016) 660–669

approach (wt = xt xt 1), where x is the original value at time t and w is the first degree differencing value at time t [63,64]. Several steps of differencing can be calculated to detrend the data. Anderson [63] noted that if the variance of an additional step of differencing increases, the sample has been over-differenced. For each variable, differencing for successive degrees was conducted until an increase in variance was observed, where the additional step was not included. The observation of standard deviations for the sample for each successive step of differencing ensured that the data was detrended to the proper degree and not overdifferenced. Table 1 presents the original values of electricity consumption, climatic, efficiency electricity savings, and costs variables as well as each successive degree a decrease in variance occurred compared to the preceding value. The study used residential survey usage data [60]. The overall kW h electricity consumption per consumer was multiplied by the percentage of electricity allocated above for heating (i.e., 15.71%) and cooling (i.e., 15.62%). IBM SPSS Statistics version 22 was used for statistical analysis. Descriptive statistics were calculated for the electricity consumption, climatic, EE savings, and EE costs variables (see Table 1). Regression models were run using stepwise linear regression. The first model examined the variability in heating kW h consumption per consumer explained by HDD, incremental (within year) and annual (accumulated) kW savings per consumer, and indirect EE costs per consumer. The second model examined the variability in cooling kW h consumption per consumer explained by CDD, incremental and annual kW savings per consumer, and indirect EE costs per consumer. HDD, CDD, and incremental kW savings per consumer used original values as the detrending increased variance after the first step. kW h per consumer for heating and cooling applications, annual kW savings per consumer, and indirect EE costs per consumer each used detrended values as decrease in variance was observed. One-way ANOVAs were utilized to examine the difference in residential energy consumer participation in EE programs based on their awareness of the programs. Likewise, One-way ANOVAs were utilized to examine the differences in sorting variables in terms of perceived utility company motives for EE programs, support for government subsidies for EE programs, and support for utility use of clean energy. Sorting variables included residential electricity consumer awareness about EE programs, previous participation in EE programs, and the demographic factors of gender, race, political party, and income. 3. Results

Table 2 Stepwise linear regression models for kW h per consumer for heating application and kW h per consumer for cooling application. Step and variables

b

SE

B

t

p

Heating kW h Step 1 (AdjR2 = .168) Indirect costs2

.458

22.36

50.157

2.24

.037*

Step 1 (AdjR2 = .168) Indirect costs2

.458

22.23

49.85

2.24

.037*

Step 2 (AdjR2 = .336) Indirect costs2 CDD0

.507 .442

55.26 .312

2.77 2.41

.013* .027*

Cooling kW h

19.977 .129

Note: Indirect costs were detrended to the 2nd degree. * p < .05.

demonstrate differences among individual groups. Research Question 2a asked if participation was significantly different for those who are aware, mis-informed, and unaware about utility efficiency programs. A significant difference was observed (F = 309.92, p < .0001), with those aware significantly more likely to participate in EE programs than mis-informed or unaware consumers, and mis-informed consumers significantly more likely to participate than unaware consumers. Research Question 2b asked if there was a significant difference in perceptions about utility motives for providing efficiency programs, support for government policy for efficiency programs, and support for utility use of clean energy based on residential electricity consumer awareness levels (see Table 3). As shown in Table 3, for each variable of interests there was a significant difference in perceptions and support based on consumer awareness at the p < .05 level. Research Question 2b also asked if there was a significant difference in perceptions about utility motives for providing efficiency programs, support for government policy for efficiency programs, and support for utility use of clean energy based on residential electricity consumer participation (see Table 4). There were significant differences for each relationship, with those who participated differing from those who had not participated and those who were unsure if they had participated. Research Question 2b asked if there were significant differences in perceptions about utility motives for providing efficiency programs, support for government policy for efficiency programs, and support for utility use of clean energy based on residential

3.1. Energy consumption, climate, and energy efficiency Stepwise linear regression was utilized to examine Research Question 1, which sought to explain variability in kW h consumption for heating and cooling applications. Indirect EE costs, or those costs not directly related to utility incentive funds, was the only significant predictor for kW h per consumer for heating applications, with the one-step model accounting for 16.8% (Adjusted R2 = .168) of the variability (Standardized b = .458, p < .001; F = 5.03, p < .05). A two-step model emerged for kW h per consumer for cooling application where indirect EE costs (Standardized b = .507, p < .05) and CDD (Standardized b = .442, p < .05) explained 33.6% (Adjusted R2 = .336; F = 6.06, p < .01) of the variability (see Table 2 for full results). 3.2. Residential energy consumer differences One-way Anovas and Scheffe’s post hoc tests were utilized for all portions of Research Question 2. Scheffe’s post hoc tests

Table 3 Relationship between awareness and residential electricity consumer perceived utility company motives, support for government subsidies for EE, and support for utility use of clean energy. Variable

Aware

Misinformed

Unaware

F

p

Utility Company Not Waste

2.06a (1.12)

2.81b (1.27)

2.90b (1.03)

75.35

.000***,*

Utility Company Save

2.17a (1.14)

3.09c (1.28)

2.87b (1.08)

77.18

.000***,*

Govt. Subsidy Support

3.24a (1.33)

3.14 (1.48)

3.29b (1.42)

3.67

.026*

Utility Clean Energy Support

2.76a (1.68)

2.52a (1.67)

2.19b (1.57)

10.34

.000***,*

Note: ⁄⁄ p < .01, ⁄⁄⁄⁄ p < .0001. Standard deviations appear in parentheses below means. Means with differing subscripts within rows are significantly different at the p < .05 based on Scheffe post hoc paired comparisons. * p < .05. *** p < .001.

C.A. Craig / Applied Energy 165 (2016) 660–669 Table 4 Relationship between participation in EE programs and residential electricity consumer perceived utility company motives, support for government subsidies for EE, and support for utility use of clean energy. Variable

Participant

Nonparticipant

Unsure

F

p

Utility Company Not Waste

2.03a (1.11)

2.83b (1.18)

2.72b (1.12)

62.67

.000***,* ***,*

665

energy (F = 2.10, p < .05). Please see Table 6 for the breakdown of income range responses to each of the four items.

4. Discussion

Utility Company Save

2.03a (1.11)

2.97b (1.22)

2.82b (1.01)

77.16

.000

Govt. Subsidy Support

2.39a (1.34)

2.84b (1.39)

2.84b (1.36)

11.76

.000****

Utility Clean Energy Support

1.91a (1.15)

2.16b (1.21)

2.37b (1.30)

8.07

.000***,*

Note: ⁄⁄ p < .01. Standard deviations appear in parentheses below means. Means with differing subscripts within rows are significantly different at the p < .05 based on Scheffe post hoc paired comparisons. * p < .05. *** p < .001. **** p < .0001.

Table 5 Relationship between race and residential electricity consumer perceived utility company motives, support for government subsidies for EE, and support for utility use of clean energy. Variable

W

B

H

O

F

p

Utility Company Not Waste

2.71 (1.19)

2.48a (1.16)

3.29 (1.25)

2.99b (1.28)

4.46

.004**

Utility Company Save

2.77a (1.24)

2.56a (1.21)

3.57 (.79)

3.30b (1.09)

7.53

.000***,*

Govt. Subsidy Support

2.86b (1.38)

2.15a (1.19)

3.71b (1.50)

3.03b (1.54)

18.72

.000****

Utility Clean Energy Support

2.17b (1.22)

1.86a (1.05)

2.43 (1.40)

2.35a (1.40)

4.98

.002**

Note: W = White, B = Black, H = Hispanic, O = Other. Standard deviations appear in parentheses below means. Means with differing subscripts within rows are significantly different at the p < .05 based on Scheffe post hoc paired comparisons. * p < .05. ** p < .01. *** p < .001. **** p < .0001.

electricity consumer demographic factors including gender, race, political party, and income as well (see Table 5). There were significant differences between males and females on utility motives for not wanting consumers to waste money (F = 10.41, p < .001), utility motives for wanting to save consumers money (F = 16.73, p < .0001), consumer support for subsidies (F = 46.56, p < .0001), and consumer support for utility use of clean energy (F = 32.654, p < .0001). In each instance, females were more likely to perceive the utility favorably, support subsidies, and support clean energy. Likewise, significant differences were found for each of the four items of interest with regards to race (see Table 6). There were significant differences between Democrats and Republicans on utility motives for not wanting consumers to waste money (F = 11.05, p < .0001), utility motives for wanting to save consumers money (F = 8.05, p < .0001), consumer support for subsidies (F = 31.62, p < .0001), and consumer support for utility use of clean energy (F = 37.62, p < .0001). In each instance, Democrats were more likely to perceive the utility favorably, support subsidies, and support clean energy. There were significant differences among income groups on utility motives for not wanting consumers to waste money (F = 3.18, p < .0001), utility motives for wanting to save consumers money (F = 2.28, p < .01), consumer support for subsidies (F = 4.41, p < .0001), and consumer support for utility use of clean

Results from Research Question 1 highlight the impact of the climatic variability experienced in the focal state, and the role that energy efficiency has on reducing residential electricity consumption. There are over $6 billion spent by utilities on each year to reduce electricity consumption [22], and in the Southeast US electricity is the primary fuel source for both heating and cooling [60]. Moreover, states in the focal region use proportionally more fossil fuels such as coal to produce electricity (EIA, 2015), intensifying greenhouse gas emissions related to electricity consumption. In order to reduce the climatic impact of residential electricity consumption in this region related to heating and cooling, it is crucial to understand the unique interactions between efficiency efforts and changing climatic characteristics. Regarding kW h consumption per consumer allocated to heating, indirect costs per consumer, including expenses such as marketing and administrative for running EE programs, was the only significant predictor. The negative relationship suggests that as spending on programs not allocated directly to incentives increases, kW h consumption per consumer for heating applications in fact decreases. Globally, mean temperatures are rising [1]. Consistent with this trend in the focal state there was a slight downward trend in HDD, an indication that there are less cold days requiring heating applications. However, kW h consumption for heating was not significantly related to HDD. CDDs were trending up during the study period in the focal state. CDD exhibited a positive relationship that resulted in an overall two-step model that also included indirect EE costs and explained 33.6% of the variation in kW h consumption for cooling. Consistent with previous research [23,37,65], our results support the notion that climate is playing a significant and important role in energy consumption and savings. Previous research suggests that communication and inclusion in efficiency efforts are crucial in reducing consumption [17,33,36,37]. There has been a rapid expansion of EE spending and communication in recent years [22]. However, in the focal state a significant decrease in actual electricity savings from efficiency efforts for heating and cooling applications was not present. Indirect costs are associated with activities such as marketing and outreach. While efforts associated with EE spending could very well be influencing actual kW h consumption related to heating and cooling behaviors, it is feasible that the data reported by utility companies may not be robust enough to clearly capture these savings. Consistent with the progression of the TPB, findings from Research Question 2a confirmed that residential electricity consumers who were aware of efficiency programs were significantly more likely to participate than those who were either unaware or who were mis-informed. A post hoc test indicated that those who were mis-informed (that is, stated there were no efficiency programs when in fact there were) were significantly more likely to participate than those who are completely unaware. When asked about participation in terms of using a rebate or discount to purchase an energy efficient CFL, the individuals who responded that there were no EE programs in fact had used significantly more rebates or discounts than those who were completely unaware. This finding demonstrates a potential shortcoming with the effectiveness of utility company communication efforts to increase participation, reduce consumption, and reduce emissions. Without rich feedback and communication related to efficiency measures [18,33,36], the trajectory of increased energy consumption will

666

C.A. Craig / Applied Energy 165 (2016) 660–669 Table 6 Relationship between income bracket and residential electricity consumer perceived utility company motives, support for government subsidies for EE, and support for utility use of clean energy.

3.5000

3.0000

2.5000

2.0000

1.5000

1.0000

0.5000

0.0000 0 - 19k 20 - 29k 30 - 39k 40 - 49k 50 - 59k 60 - 69k 70 - 79k 80 - 89k 90 - 99k 100 124k Utility - No Waste

Utility - Save

likely continue, and the additional benefits of diffusion of socially conscious activities will remain unrealized [45,46]. Further, the TPB states that awareness and knowledge help to build positive attitudes and perceptions towards a topic. Participation without awareness bi-passes this step, decreasing the potential for positive attitudes, perceptions, and future behaviors to occur [43]. Research Question 2b asked if there was a significant difference in perceptions about utility motives for providing efficiency programs, support for government policy for efficiency programs, and support for utility use of clean energy based on residential electricity consumer awareness levels, participation in utility programs, and demographic factors. Those who were aware of utility EE programs were significantly more likely to perceive that utilities offered programs as to help the residential consumer not waste money. With regards to the utility company’s motives being to save the consumer money, those who were aware were significantly more likely to favorably view utility motives than the unaware or mis-informed individual, and the unaware individual was significantly more likely than the mis-informed individual to view motives favorably. When considering the relationship between awareness and participation addressed with Research Question 2a, these findings are not favorable for utility companies. While the mis-informed consumer is not as likely as the informed to participate, they are more likely than the uniformed to participate, and they are the most likely group to negatively view utility motives for EE programs. In keeping with the TPB, misinformation is more strongly related to negative attitudes than no information at all. Consistent with Pelletier and Sharp [66], positive attitudes and perceptions appeared to follow knowledge and literacy associated with awareness about efficiency programs. Kim [67] states it takes more than awareness for the enactment of pro-environmental actions such as support for policy or utility company use of clean energy. With the TPB awareness is the starting point to build positive attitudes and perceptions, and to move towards action. The findings of the study support previous research (e.g., [18,19]) that awareness remains low, making it less likely on the aggregate the pro-conservation attitudes or behaviors will occur. A slightly different picture emerged with regards to government subsidies for EE and utility use of clean energy. Aware

Subsidy

125k - 150k - 250k + 149k 249k

Clean Energy

consumers were only significantly more likely than unaware consumers to be more supportive of subsidies for EE, and both the aware and mis-informed consumers were significantly more likely than the unaware consumers to support utility use of clean or alternative energy. Considering that mis-informed consumers trust utility motives less than the other groups, it is not surprising that they are not supportive of subsidies for utility companies, even if they are more likely than the unaware to participate in efficiency programs. However, the mis-informed consumer is more similar to the aware consumer with support for utility use of clean energy. The reasoning behind this support may vary, however. It stands to reason since mis-informed consumers are more likely to view utility companies less favorably than other groups, that the support for the use of clean energy could be related to holding the utility company accountable. For Research Question 2c, those who participated in utility EE programs were significantly different from both the unaware and the mis-informed for all independent variables of interest. Keeping in mind those who were aware of efficiency programs were more likely than the mis-informed or unaware, these findings are consistent when viewed through the lens of the TPB. Consistent with past research (e.g., [18,19,50]), significant differences based on demographic factors emerged. The differences between women and men and Democrats and Republicans were the most pronounced, with women and Democrats perceiving the utility company more favorably on both items and being more supportive of both subsidies and utility use of clean energy. This supports previous literature that shows women and Democrats are more supportive pro-environmental initiatives than men (e.g., [18,49,51]). Regarding race, consumers who self-identified as Black and Other were significantly more supportive of both subsidies for utility efficiency programs and utility use of clean energy than consumers who self-identified as White. As shown in Table 6, there were significant differences based on income as well for both utility motive independent variables and support of government subsidies. Post-hoc tests indicated that there were only two groups that were significantly different from one another, residents in the 0–19 K income bracket and residents in the 100–124 K income bracket. Residents in the lowest income bracket (0–19 K) perceived utility motives more favorably than

C.A. Craig / Applied Energy 165 (2016) 660–669

the 100–124 K income bracket, and were more supportive of government subsidies. Specific to subsidies for utility bill assistance, the federal government established the need-based Low Income Housing Assistance program (LIHEAP) in 1981 to help with heating and cooling expenses to avoid service disconnect, to respond to extreme weather events that interrupt service, to help with lowcost home upgrade projects to lower demand, and to provide consulting services to help reduce consumption [68]. According to Perl $3.615 billion were appropriated to fund LIHEAP in 2014, and in 2009 an estimated 35 million low-income households were eligible to participate. It may be that higher awareness levels and participation in government assistance or subsidy programs such as LIHEAP are similar to awareness and participation to EE programs in the current study, leading to higher levels of support for policy. One of the most interesting findings was with the highest income bracket (250 K +). While not the most supportive income bracket, this group was more consistent with the lower income brackets than the higher income brackets, indicating that while the extremely wealthy may not perceive utility motives more favorably, they are in fact more supportive of energy and emissions reduction through efficiency subsidies and clean energy infrastructure.

667

resource availability issues. There are billions of dollars [22] being deployed for EE programs, however, actual, verifiable electricity savings were not significantly related to the residential kW h consumption associated with heating or cooling. Delmas et al. [33] and Darby [36] noted that the energy savings are closely related to feedback and communication. Those who are aware are more likely to participate and have positive views about the utility company’s motives, which in turn could help to expand socially conscious behaviors to other consumers through normative pressures or nudges [45,46]. However, despite residential consumers paying into the EE programs on every month’s utility bill, the vast majority are not aware or are mis-informed, and have not participated. This raises a serious question about the motives of the utility companies for the programs, and the manner in which spending is occurring as to engage residential consumers in true and transformational change to become more efficient. Also, by utilizing a method that explicitly takes into account climatic variability and EE programs, on a macro-scale this information could be used to more accurately forecast the supply and demand of energy to help answer difficult energy security and policy issues. 4.3. Limitations and future research

4.1. Conclusions The results of the current study demonstrate that climatic variability can play a major role in kW h consumption among residential electricity consumers, particularly in areas that are reliant on electricity for heating and cooling homes. Bradshaw et al. [65] note the interactive nature of climatic conditions with regards to efficiency. Fisher and Newell [21] discuss the need for both policy (e.g., utility EE programs, electricity pricing, energy mix) as well as the diffusion of knowledge to deploy environmental technologies and processes. In the focal state, policy continues to expand the spending on utility EE spending [22]. However, the lack of actual, verifiable electricity savings related to heating and cooling suggests that communication, as well as the diffusion of technologies and processes related to pro-conservationism, can be improved when targeting residential consumers. The current study sought to lay out a clear path using the TPB that can move residential electricity consumers from awareness about utility programs to participation in programs. Furthermore, the study sought to draw the connection between awareness, perceptions about utility motives, and consumer support for government subsidies and utility clean energy use. While not causally related, the study did show that those who were aware of EE programs had more favorable perceptions about utility motives, were more likely to participate in programs, and were more likely to support subsidies and clean energy use by utilities. The research offered an oft-utilized approach that examined the interaction between the climate, the environment in terms of energy consumption, and residential electricity consumers. The findings in the focal Southeast US state are very telling, providing support that climate is the driving influencer of kW h consumption related to cooling and that indirect EE spending is influencing kW h consumption related to both heating and cooling. Finally, people who are aware of efficiency programs are more likely to participate, perceive utility motives more favorably, and maybe most importantly, support consumption reduction and emissions reductions steps by governmental entities and utility companies to help combat the anthropogenic induced climatic variability and change [1]. 4.2. Practical implications The current study demonstrated the need for effective communication between organizations and those at the micro-level when deploying efficiency programs to help mitigate climatic and

The current study is not without limitation. First, reporting for energy consumption, EE savings, and costs was annualized. Steps were taken to utilize previously collected residential survey information about consumptive habits in order to combat this shortcoming. Next, there were a limited number of years of efficiency kW and kW h consumption data that could be matched with the HDD and CDD data. In many states, however, spending has been limited until recently with regards to efficiency programs, so data available to analyze is limited. Also, the energy data was aggregated at the state-level due to utility company reporting practices. With the availability of 4 km resolution climatic data back to 1981, the relationships between climate, consumption, and efficiency could be much better understood with location data for residences which is currently unavailable on a macro scale. The humanistic-focused portion of the study was exploratory in nature, seeking to more clearly understand differences in ratepayer awareness, perceptions, and behaviors. The large sample size (n = 2450) helped to overcome validity issues related to study limitations that included: the use of cross-sectional data, the lack of pre- and post-test design, the use of a phone survey, the higher ratio of Democrat respondents, the higher ratio of female respondents, the older age of the sample, and the use of nominal level data for awareness and participation items. States in regions outside the Southeast US are not as reliant on electricity for heating and cooling [60], and are also unique in efficiency policies and deployment down to the state-level [22]. Future research could expand to the effectiveness of efficiency programs at the state-level when controlling for climatic variability for all of the lower 48 states. Also, climate modeling could be utilized to predict trends in HDD and CDD to more accurately predict electricity consumption on a state-by-state basis when accounting for the effectiveness of current efficiency programs and spending. From a regulatory stand-point, future research could be conducted to examine the effectiveness of ‘‘deemed” savings versus actually reported savings. The current study demonstrates significant differences in pro-environmental attitudes, support for EE subsidies, and support for clean energy. Future research should be expanded nationally in the US to determine the driving factors for residential EE support and clean energy use by utility companies among residential electricity consumers both within similar climate regions and on a state-by-state basis. Researchers have a unique opportunity to move forward by integrating change agents (e.g., residential electricity consumers) into climate and energy studies as to

668

C.A. Craig / Applied Energy 165 (2016) 660–669

increase the likelihood of the enactment of positive change to mitigate pressing issues such as anthropogenic forced climate change and resource availability.

[29]

[30]

References [1] IPCC. Climate change 2014: mitigation of climate change. In: Edenhofer O, Pichs-Madruga R, Sokona Y, Farahani E, Kadner S, Seyboth K, et al., editors. Contribution of working group III to the fifth assessment report of the intergovernmental panel on climate change. Cambridge (United Kingdom and New York, NY, USA): Cambridge University Press; 2014. [2] United States Energy Information Agency (EIA). International energy statistics. United States Department of Energy; 2015. [retrieved 10/10/2015]. [3] Heede R. Tracing anthropogenic carbon dioxide and methane emissions to fossil fuel and cement producers, 1854–2010. Climate Change 2014;122:229–41. [4] United States Energy Information Agency (EIA). Retail sales of electricity by state by sector. United States Department of Energy; 2015. [retrieved 10/10/2015]. [5] Wang M, Tian L. Regulating effect of the energy market – theoretical and empirical analysis based on a novel energy prices–energy supply–economic growth dynamic. Appl Energy 2015;155:526–46. [6] Zhao X, Yin H, Zhao Y. Impact of environmental regulations on the efficiency and CO2 emissions of power plants in China. Appl Energy 2015;149:238–47. [7] Haerer D, Pratson L. Employment trends in the U.S. electricity sector, 2008– 2012. Energy Policy 2015;82:85–98. [8] Ryu H, Dorjragchaa S, Kim Y, Kim K. Electricity-generation mix considering energy security and carbon emission mitigation: case of Korea and Mongolia. Energy 2014;64:2014. [9] Karanfil F, Li Y. Electricity consumption and economic growth: exploring panel-specific differences. Energy Policy 2015;82:264–77. [10] Quadrdan M, Chaudry M, Jenkins N, Barauh P, Eyre N. Impact of transition to a low carbon power system on the GB gas network. Appl Energy 2015;151:1–12. [11] Schill W, Clemens G. Power system impacts of electric vehicles in Germany: charging with coal or renewables. Appl Energy 2015;156:185–96. [12] Shahiduzzaman M, Layton A. Changes in CO2 emissions over business cycle recessions and expansions in the United States: a decomposition analysis. Appl Anal 2015;150:25–35. [13] Brouwer AS, van den Broek M, Seebregts A, Faaij A. Operational flexibility and economics of power plants in future low-carbon power systems. Appl Energy 2015;156:107–28. [14] Hanak DP, Biliyok C, Manovic V. Efficiency improvement for the coal-fired power plant retrofit with CO2 capture plant using chilled ammonia process. Appl Energy 2015;151:259–72. [15] Sanna A, Ramli I, Maroto-Valer MM. Development of sodium/lithium/fly ash sorbents for high temperature post-combustion CO2 capture. Appl Energy 2015;156:197–206. [16] Hanimann R, Vinterback J, Mark-Herbert C. Consumer behavior in renewable electricity: can branding in accordance with identity signaling increase demand for renewable electricity and strengthen supplier brands? Energy Policy 2015;78:11–21. [17] Liu W, Zhang J, Bluemling B, Mol AP, Wang C. Public participation in energy saving retrofitting of residential buildings in China. Appl Energy 2015;147:287–96. [18] Craig CA, Allen MW. Enhanced understanding of understanding energy ratepayers: factors influencing perceptions of government energy efficiency subsidies and utility alternative use. Energy Policy 2014;66:224–33. [19] Langevin J, Gurian PL, Wen J. Reducing energy consumption in low income public housing: interviewing residents about energy behaviors. Appl Energy 2013;102:1358–70. [20] Shove E. Beyond the ABC: climate change policy and theories of social change. Environ Plann A 2010;42:1273–85. [21] Fisher C, Newell RG. Environmental and technology policies for climate mitigation. J Environ Econ Manage 2008;55:142–52. [22] Gilleo A, Chittum A, Farley K, Neubauer M, Nowak S, Ribeiro D, et al. The 2014 state energy efficiency scorecard. American Council for an Energy-Efficient Economy; 2014. . [23] Jovanovic S, Savic S, Bojic M, Djordjevic Z, Nikolic D. The impact of the mean daily air temperature change on electricity consumption. Energy 2015;88:604–9. [24] Diaz HF, Quayle RG. Heating degree day data applied to residential heating energy consumption. J Appl Meteorol 1980;3:241–6. [25] Timmer RP, Lamb PJ. Relations between temperature and residential natural gas consumption in the Central and Eastern United States. J Appl Meteor Climatol 2007;46:1993–2013. http://dx.doi.org/10.1175/2007JAMC1552.1. [26] United States Environmental Protection Agency (EPA). Heating and cooling degree days. Environmental protection agency, technical documentation; 2014. [revised January 2014, retrieved 1/21/2015]. [27] Mourshed M. Relationship between mean annual temperature and degreedays. Energy Build 2012;54:418–25. [28] McFarland J, Zhou Y, Clarke L, Sullivan P, Colman J, Jaglom WS, et al. Impacts of rising air temperatures and emissions mitigation on electricity demand and

[31]

[32]

[33]

[34] [35] [36]

[37] [38]

[39] [40] [41] [42]

[43] [44] [45] [46] [47]

[48] [49]

[50] [51]

[52]

[53]

[54] [55]

[56]

[57]

[58]

[59]

supply in the United States: a multi-model comparison. Climate Change 2015;131:111–25. Hou Y-L, Mu H-Z, Dong G-T, Jun S. Influences of urban temperature on the electricity consumption of Shanghai. Adv Clim Change Res 2014;5. http://dx. doi.org/10.3724/SP.J.1248.2014.074. Schurr A, Hauser S. Carpe diem – Why retail electricity pricing must change now. In: Sionshansi FP, editor. Energy Efficiency: Towards the End of Demand Growth. Oxford, UK: Academic Press; 2013. p. 111–32. Ingram K, Dow K, Carter L, Anderson J. Climate of the Southeast United States: variability, change, impacts, and vulnerability. Washington, DC: Island Press; 2013. Sioshansi FP. Will energy efficiency make a difference? In: Sionshansi FP, editor. Energy efficiency: towards the end of demand growth. Oxford (UK): Academic Press; 2013. p. 3–49. Delmas MA, Fischlein M, Asensio OI. Information strategies and energy conservation behavior: a meta-analysis of experimental studies from 1975 to 2012. Energy Policy 2013;61:729–30. Gillingham K, Kotchen MJ, Rapson DS, Wagner G. Energy policy: the rebound effect is overplayed. Nature 2013;493:475–6. Asensio OI, Delmas MA. Nonprice incentives and energy conservation. PNAS; 2015 [published online January 12, 2015]. Darby S. The effectiveness of feedback on energy consumption: a review for DEFRA of the literature on metering, billing and direct displays. Environmental Change Institute, University of Oxford; 2006. Craig CA, Allen MW. The impact of curriculum-based learning on environmental literacy, energy consumption, and policy. Utilities Policy 2015;35:41–9. DeWaters JE, Powers SE. Energy literacy of secondary students in New York State (USE): a measure of knowledge, affect, and behavior. Energy Policy 2011;39(3):1699–710. Allen MW. Strategic communication for sustainable organizations – theory and practice. New York (NY): Springer; forthcoming. Gupta S, Ogden DT. To buy or not to buy? A social dilemma perspective on green buying. J Consum Mark 2009;26:376–91. Pickett-Baker J, Ozaki R. Pro-environmental products: marketing influence on consumer purchase decision. J Consum Mark 2008;25:281–93. Unsworth KL, Dmitrieva A, Adriasola E. Changing behavior: increasing the effectiveness of workplace interventions in creating pro-environmental behaviour change. J Organ Behav 2013;34:211–29. Ajzen I. The theory of planned behavior. Organ Behav Hum Decis Proc 1991;50:179–211. Craig CA, Allen MW. Sustainability information sources: employee knowledge, perceptions, and learning. J Commun Manage 2013;17:292–307. Cialdini RB. Crafting normative messages to protect the environment. Psychol Sci 2003;12:105–9. Thaler RH, Sunstein CR. Nudge: improving decisions about health, wealth, and happiness. New Haven (CT): Yale University Press; 2008. Lulfs R, Hahn R. Sustainable behavior in the business sphere: a comprehensive overview of the explanatory power of psychological models. Organ Environ 2014;27:43–64. Nilsson A, Andersson K, Bergstad CJ. Energy behaviors at the office: an intervention study on the use of equipment. Appl Energy 2015;146:434–41. Aktamis H. Determining energy saving behavior and energy awareness of secondary school students according to socio-demographic characteristics. Educ Res Rev 2011;6:243–50. Brouhle K, Khanna M. Determinants of participation versus consumption in Nordic Swan eco-labeled market. Ecol Econ 2012;73:132–51. Coffey DC, Joseph PH. A polarized environment: the effect of partisanship and ideological values on individual recycling and conservation behavior. Am Behav Sci 2013;57:116–39. United States Environmental Protection Agency (EPA). Overview of the clean power plan: cutting carbon pollution from power plants. Environmental protection agency fact sheet: clean power plan; 2014. [retrieved 4/25/2015]. Molina M. The best value for America’s energy dollar: a national review of the cost of utility energy efficiency programs. American Council for an energy efficient economy; 2014. [retrieved 11/30/2015]. Trochim WM. The research methods knowledge base. Cincinnati: Atomic Dog Publishing; 2001. Xing Y, Handy S. On-line vs. phone surveys: comparison of results for a bicycle survey. Institute of Transportation Studies, University of California, Davis; 2011 [retrieved October 10, 2013]. PewResearch Center for the People & the Press; 2012. [retrieved October 10, 2013]. United States Census Bureau. State & county quick facts; 2015. [retrieved 12/14/ 2015]. Shelton Group. Full spectrum insights: a new way to motivate sustainable behaviors; 2011. [retrieved October 10, 2013]. United States Energy Information Agency (EIA). Residential energy consumption survey (RECs); 2009a. [retrieved 3/15/2015].

C.A. Craig / Applied Energy 165 (2016) 660–669 [60] United States Energy Information Agency (EIA). State fact sheets on household energy use; 2009b. [retrieved 3/15/2015]. [61] National appliance energy conservation act of 1987, 100 U. S. C. . [62] United States Department of Energy (DOE). Furnaces and boiler; 2015. [retrieved 4/27/2015]. [63] Anderson O. Time series analysis and forecasting: the box-jenkins approach. London: Butterworths; 1976. [64] Yaffee RA, McGee M. An introduction to time series analysis and forecasting: with applications of SASÒ and SPSSÒ. Academic Press; 2000. [65] Bradshaw JL, Bou-Zeid E, Harris RH. Comparing the effectiveness of weatherization treatment for low-income, American, urban housing stocks in different climates. Energy Build 2014;69:535–43.

669

[66] Pelletier LG, Sharp E. Persuasive communication and proenvironmental behaviours: how message tailoring and message framing can improve the integration of behaviors through self-determined motivation. Can Psychol 2008;49:210–6. [67] Kim Y. Understanding green purchase: the influence of collectivism, personal values, and environmental attitudes, and the moderating effect of perceived consumer effectiveness. Seol J Bus 2011;17:65–92. [68] Perl L. LIHEAP: program and funding. Congressional research service, RL31865 (7-5700) July 18, 2013; 2013. [retrieved 10/8/2015].